Well, i completelly took appart a mode hopping DX200, and observed something interesting.
There is no IR filter inside, and from the pictures or others taken appart, neither is there one in the DX100 and True DX100, but i think they don't need one.
People have measured low IR and thought it means a mediocre filter, but i believe the reason is somewhat different.
If i take a picture of the crystals, without the expander, i see a LOT of IR glow around the green beam. With the expander on, this glow decreases a little, at the collimator it decreases even more, and with the top on, the IR glow is gone.
I also noticed, that if i misalign the crystalls, the amount of the IR leaking increases, so i did this and again repeated the procedure. There was noticably more IR at the lower stages, but at the top, it was gone again.
So i believe, the reason for low IR leaking is this:
The IR isn't collimated, before it enters the crystals, but the green comes out very narrow.
The IR spreads out, when it leaves the crystals, and there is a narrow expander lens, which expands the green, and some IR, but blocks most of the IR.
At the collimator, there is again some IR blocked, and mostly only the green collimated.
At the top of the laser, the narrow apperture blocks even more IR.
At each of these stages, the amount of the IR decreases, and at the top, there is very little left.
Higher quality green lasers have a collimator between the pump diode and the crystal assembly (or at least a fast axis lens), and maybe this is the reason, they would leak more IR, if there wasn't a filter.
My thermopile hasn't arrived yet, but i haven't finished putting the laser back together, because i need the meter for final finetuning, so i will repeat this process in front of the thermopile, with a filter, that will only allow IR to pass.